Capabilibara - Capability Provenance in Language Models
🦫
Capability provenance in language models (COLM 2026).
AI & ML interests
None defined yet.
Recent Activity
View all activity
glennmatlin
updated a
Space about 4 hours ago
glennmatlin
published a
Space about 4 hours ago
glennmatlin
updated 2
buckets 4 days ago
glennmatlin
published 3
buckets 4 days ago
glennmatlin
published a
bucket 5 days ago
glennmatlin
updated 2
buckets 5 days ago
glennmatlin
updated a
bucket 8 days ago
glennmatlin
updated a
dataset 23 days ago
glennmatlin
updated a
bucket 26 days ago
glennmatlin
published a
bucket 26 days ago
glennmatlin
updated a
bucket 26 days ago
glennmatlin
published a
bucket 26 days ago
Akhil-Theerthala
submitted a
paper to Daily Papers 7 months ago
Akhil-Theerthala
authored a
paper 7 months ago
Akhil-Theerthala
posted an update 7 months ago
Post
511
Is it better to show a model too many images once (Diversity), or extract as much information from a small set of images?
I have always wanted to do an ablation study on this and recently I got the chance to do exactly that. Why? In applied domains like robotics, manufacturing, or banking, we rarely have the luxury of internet-scale diverse image datasets. We are often "Data Poor" in terms of diversity but "Data Rich" in depth.
The takeaway? Density is efficient for facts but dangerous for reasoning (logical collapse) if you don't have larger scale data.
More details:
https://huggingface.co/blog/Akhil-Theerthala/diversity-density-for-vision-language-models
I have always wanted to do an ablation study on this and recently I got the chance to do exactly that. Why? In applied domains like robotics, manufacturing, or banking, we rarely have the luxury of internet-scale diverse image datasets. We are often "Data Poor" in terms of diversity but "Data Rich" in depth.
The takeaway? Density is efficient for facts but dangerous for reasoning (logical collapse) if you don't have larger scale data.
More details:
https://huggingface.co/blog/Akhil-Theerthala/diversity-density-for-vision-language-models
- 1 reply
Akhil-Theerthala
posted an update 9 months ago
Post
1368
For anyone who likes to get their papers summarized, I've tried to make a bentocard summarizer for research papers, that enables us to get overviews of the papers, with possible deepdives where interested. Of course, this has to be improved a lot, so do check the space and give your feedback!
Akhil-Theerthala/PaperStack
Some sample bentogrids are as follows:
Akhil-Theerthala/PaperStack
Some sample bentogrids are as follows: